Vertical faceting

The question is: how can we plot one variable for faceting vertically?

And bonus, since we want through time the plot of very different measurements, we must have a free y-scale.

TL;DR

Use either:

• The dir = "v" argument of facet_wrap()
• The rows = vars(variable) argument of facet_grid()

Demo

We will use a file from the Aranet4 sensor using the mobile app provided by the company. This sensor measures 4 parameters every 2 to 10 minutes which are:

• CO₂, using the infrared absorption property of greenhouse gases like carbon dioxide.

• Temperature in Celsius degrees.

• Relative humidity (in %).

• Atmospheric pressure in _h Pa_. This latter variable is exported but displayed on the device, we will skip it.

The CO₂ measurement is performed as follows (from the user manual):

CO2 aranet

According to the file extension, it is a comma-separated values. But if you look at the file with a text editor, you will see that values are surrounded by double quotes and number are using commas for decimal separator. Moreover, the dates are not in a standard format unfortunately. But it is a real life example data.

readr will detect the double quotes and deal with it automatically but the other issues needs to be fixed:

• For the Time(dd/mm/yyyy) column, the datetime format is specified.
• For the number decimals as commas, you need to set the locale to locale(decimal_mark = ",")
• Skip the last column (Atmospheric Pressure) as we won’t plot it.

aranet <- read_csv("https://biostat2.uni.lu/practicals/data/Gino_2022-08-14T22_53_06+0200.csv",
                   col_types = cols(`Time(dd/mm/yyyy)` = col_datetime(format = "%d/%m/%Y %H:%M:%S")),
                   locale = locale(decimal_mark = ","),
                   col_select = -`Atmospheric pressure(hPa)`) |> 
  rename(Time = `Time(dd/mm/yyyy)`)
aranet

# A tibble: 4,033 × 4
   Time                `Carbon dioxide(ppm)` `Temperature(°C)`
   <dttm>                              <dbl>             <dbl>
 1 2022-07-31 22:53:43                   457              24  
 2 2022-07-31 22:58:43                   444              24.1
 3 2022-07-31 23:03:43                   457              24  
 4 2022-07-31 23:08:43                   444              24  
 5 2022-07-31 23:13:43                   445              24  
 6 2022-07-31 23:18:43                   454              24.1
 7 2022-07-31 23:23:43                   432              24.1
 8 2022-07-31 23:28:43                   442              24.1
 9 2022-07-31 23:33:43                   469              24  
10 2022-07-31 23:38:43                   455              24  
# ℹ 4,023 more rows
# ℹ 1 more variable: `Relative humidity(%)` <dbl>

Reshape the variables in the long format

aranet |>
  pivot_longer(cols = -Time,
               names_to = "measure",
               values_to = "value") -> aranet_long
aranet_long

# A tibble: 12,099 × 3
   Time                measure              value
   <dttm>              <chr>                <dbl>
 1 2022-07-31 22:53:43 Carbon dioxide(ppm)  457  
 2 2022-07-31 22:53:43 Temperature(°C)       24  
 3 2022-07-31 22:53:43 Relative humidity(%)  49  
 4 2022-07-31 22:58:43 Carbon dioxide(ppm)  444  
 5 2022-07-31 22:58:43 Temperature(°C)       24.1
 6 2022-07-31 22:58:43 Relative humidity(%)  49  
 7 2022-07-31 23:03:43 Carbon dioxide(ppm)  457  
 8 2022-07-31 23:03:43 Temperature(°C)       24  
 9 2022-07-31 23:03:43 Relative humidity(%)  49  
10 2022-07-31 23:08:43 Carbon dioxide(ppm)  444  
# ℹ 12,089 more rows

Plot the values per Time and facet per measurement

Measurement are one of the 3 variables: CO₂ ppm, Temp and Relative Humidity.

In the facet command, you should free the y-axis with scales = "free_y"

We want variables in rows. Either facet_wrap() dir argument or facet_grid() rows = do the job.

facet_wrap() solution

The key part is to change the dir argument for vertical

ggplot(aranet_long, aes(x = Time, y = value)) +
  geom_line() +
  facet_wrap(vars(measure), scales = "free_y",
             # panel titles next to the y axis
             strip.position = "bottom", # use left for mimicking facet_grid below
             dir = "v") +
  scale_x_datetime(breaks = scales::date_breaks("1 day"),
                   date_labels = "%b %d") +
  labs(x = NULL, y = NULL) +
  theme_minimal(14)

facet_grid() solution

faacet_grid() is designed for two variables faceting, but we can omit one.

The key part is to use only the `rows`` argument for vertical panels.

ggplot(aranet_long, aes(x = Time, y = value)) +
  geom_line() +
  facet_grid(rows = vars(measure), scales = "free_y",
             # panel titles next to the y axis
             switch = "y") +
  scale_x_datetime(breaks = scales::date_breaks("1 day"),
                   date_labels = "%b %d") +
  labs(x = NULL, y = NULL) +
  theme_minimal(14)